I started with the Hurley method when I first started working on my autotrader. I made a WHOLE bunch of changes since then and you wouldn't even recognze the two if you laid them side by side. The biggest issue I had with that mehtod was the way the code was laid out. I am a real "order" kind of person and found it difficult to follow the code in some places. Then with everybody adding their improvements to the code, it was near impossible to keep up. I didn't like the thought of just taking someone's strategy and using it without knowing what was under the hood. [It's a control thing, I guess.]
I don't know if this helps or not. I will tell you that working on an autotrader is not for the weak of heart. I think the only thing that helped me through was the support group I had. There are three other guys that I worked with in developing my code. They use the code, debug it, encourage me, listen to the griping, do back testing, offer suggestions for improvement, pressurre me to get things working, and just generally make me feel better about the project. Without that small network I would not have made the autotrader work. It was too long of a process and too hard in several places.
So you may ask why I didn't use my friends here at Big Mike's. The problem is all me. I am a bit of a control freak and I don't like people touching my code. It's not that they don't know things, it's just that I have this way that I code that makes me feel in charge. Stupid little stuff like how I name my variables. For example, I will initialize a variable like macdSmoothing. Then when I put it through the Parameters region I will use MacdSmoothing. Some people use _macdSmoothing, or MDS or Smooth...they are all fine but it makes my head hurt and my NEED for order in the universe so I am not in conflict. I also like the use of #region and use it to infinitum (sp). Wierd right? My wife thinks so. :-)
I don't mind sharing as you might have noticed in the download section, I just don't like people touching my stuff...isn't that crazy?
Anyway, long answer to a short question. I am attaching an older version of my first cuts. The reason I am putting this one here is it is pretty basic and doesn't have all the biasis I have added over the last six months. Feel free to hack away at this. It is basic enough that I think you can put your strategy in here and see if it works. I would be more than happy to help you through the process if you decide to do this, but I must caution you...this is HARD! Good luck.
The following 2 users say Thank You to gordo for this post:
Well, let’s put this one in the books. Nice day on the ‘floor’ today. I traded a new autotrader strategy my colleagues developed over the weekend and it did nicely. Our results between the three of us (one MIA) was around $2,000 each. Not bad. Here’s a picture of one of the trades (attached - Trade1).
We are using this setup (attached-1.txt).
Last week’s test was a disaster from a monitary point of view. Here are the results (attached - Screen1).
There is a whole lot of blood on those results…lots of blood. I do find it interesting though, that the four of us using the exact same setups and the exact same program vary in results so widely. Very interesting.
I also traded one contract manually. I lost my laptop (R.I.P. 2011) Friday so I am in the process of replacing it. Therefore, I do not have my results chart. I made 12 trades (a few more than I wanted of course) and netted $219.76 after commission. I was using my indicators but watching price action more than anything else. It was nice feeling. It is really hard to go back and look at a chart afterwards and see the indicators and get the ‘feel’ for price action. It is just not the same. How could I bottle that feeling…hmmmmm… Anyway, here's a shot of my trading screen (attached - Screen2).
Looking forward to another interesting day tomorrow!
The following 3 users say Thank You to gordo for this post:
Gordo - wanted to thank you for your thread (and your apparently massive contributions to this forum in general).
I recall a time early in my career when I was unemployed and had a very difficult time finding new employment. The toll on me mentally was high and I was only married no kids. I can only imagine being in a similar situation but with kids/wife/gkids etc. That mental stress has got to be alot to overcome.....
One thing i have noticed on my successful days/trades is that I am completely calm/serene. My bad days - quite the opposite. In fact I now have a rule that the first trade i take that violates any of my trade rules ENDS THE DAY FOR ME. I have recently begun to focus on pure mental side - the real touchy feely stuff.
if you havent google "the Secrets to Emotion Free Trading"...it is a free book out there. worth reading even if you dont use it. If you cant find it just PM me an i can email it or fax it to you (i only have some hard copy i found online and printed).
The following 3 users say Thank You to greenroomhoo for this post:
Looks like I am now a statistic. I couldn't make it happen. I am back in the job market and hopefully soon of one of the employed. On the one hand, I am sad that I failed. On the other hand, I learned a great deal. So I say thank you to all of those who joined me over the years and hope you have better fortune than myself.
Now as far as this blog is concerned, I want those of you reading to know that I am now writing for me. I am not looking for back patting or sad faces. I am just using this wonderful forum as a place to put my thoughts as they come and go.
I started a website (www.midwestindicator.com). I still have lots to do, but I think I have something that is worth something to someone. We'll see. In particular, I have an auto trader that works pretty nice. I have three friends beta testing for me and it works really well. Nothing spectacular, just nice. We've been running for about six months now. Still a work in progress. Unfortunately, my programming time will be curtailed because of having to bring home the bacon.
I am learning a new trick. What was that about old dogs????? I'm still a pup with a whole life of wild experiences. My youngest son isn't so sure about some of them...little does he know...little does he know...
My new trick? Neural networks. I know there are those out there that will tell me that I am out of my tree. But it's a new tree. It is not an easy tree either. I am having to teach myself Visual Studio. I took a really great course at: http://www.learnvisualstudio.net/ . It cost me $75 but is well worth the money. If you want to learn C#, take a look.
Then I needed learn about neural networks. Now you talk about a great site...this should go on the list: http://www.heatonresearch.com/ . He has some books for sale, but most of the stuff is free. He has some GREAT You Tube videos that do some great explaining. The problem with the site is the technology. It is truly BLEEDING EDGE stuff. It is not for green horns. The examples and texts lag the software because the software is moving forward at the speed of light! So if you are looking for a place to have your hand held through the process, don't hit the link above. You will be sad. There will be tears.
So I am working through the examples. Some good, some not so good. I have now succeeded to export data, build a neural network model and then reload the results back into Ninja Trader. It is not great, but I was successful. Now I have to thank a new friend I met on the site from Morocco, France. He spent two hours walking though some of the nuances. It was great. Good thing he speaks French, cause I only speak English and BAD English. I have to type my posts in Word and copy them over, because my typing is so bad. That’s why I like to program…the compiler ALWAYS checks my spelling.
So here’s my thought. I can use the outputs out of my indicators I built (I have six sets in mind) and build a neural network. Then I can use that to give me an indicator I can use in my auto trader. That’s my plan. After all, I have a few more years in me, and the market is gonna be there.
So I move on. Best of luck to all of you! Happy trading.
The following 5 users say Thank You to gordo for this post:
So I am progressing with my implementation, but things are not as rosy as I would hope. I am confused about the results I obtained and searching for answers. I am trying to get Encog to predict where the macd will go. I picked this because the sparce examples available seem to go down this path. This is truly bleeding edge stuff.
From Ninja Trader I created a macd.csv output file to load into the Encog Workbench. I followed the instructions for version 3.0.1 in the newest article. From this I got an Encog Workbench output (below) which compared nicely with the macd_avg. Here is a partial look at the macd_output file:
(see attachment 1)
My original data which included the date_time, macd_avg, macd and macd_diff came from the following code. I only included the pertinenet parts of the code for readability.
(see attachment 2)
When I implemented the trained macd file (macd_train.eg) back into the chart I was disappointed with the results. From the historical data, I thought I would get the same answers that I got from the segregated evaluation data shown above. After all, it is exactly the same inputs with the same weights etc. They are kind of close, but since it was the exact same data, I felt I should get the exact same answer. Here are the three columns to compare:
(see attachment 3)
The macd column is the original macd value. The workbench column is the values I got from the workbench macd_output.csv file. The ninja column is the values Encog predicted inside Ninja Trader and the difference column is the difference between the workbench and ninja columns (which I thought should be identical).
I also included a chart showing that things are 'similar' but not the same....
(see attachment 4)
Here is the code I used to implement the Ninja side of the Encog program. (Once again, I only included the significant portion of the code):
(see attachment 5)
Now I am just trying to figure out why they don't jive.
Last edited by gordo; September 25th, 2011 at 10:55 PM.
The following user says Thank You to gordo for this post: